Level Analysis
Janette Avelar
5/23/2022
Cycle 5 (6.17.22)
Guiding questions/tasks & format:
SCHOOL TYPE: How do tagging patterns differ among district,
charter, independent, sws, virtual, microschool, homeschool, and hub
schools?
* table of biggest differences between categories
* odds ratio
* visualization
Analysis Notes:
Independent cases * district = 78
* charter = 57
* independent = 20
* sws = 5
* virtual = 9
* microschool = 12
* homeschool = 5
* hub = 3
Multicoded cases * district AND charter = 4
* district AND sws = 5
* district AND virtual = 7
* district AND microschool = 1
* district AND homeschool = 1
* charter AND sws = 1
* charter AND virtual = 2
* charter AND hub = 1
* independent AND virtual = 1
* independent AND microschool = 7
* independent AND hub = 1
* sws AND virtual = 1
* sws AND microschool = 1
* virtual AND microschool = 1
* virutal AND homeschool = 1
* virtual AND hub = 1
* microschool AND homeschool = 1
* microschool AND hub = 1 * NONE = 5
* Boston Day And Evening Academy
* CodeRVA Regional High School
* The Metropolitan Regional Career And Technical Center
* Grant Beacon Middle School
* Utah Schools For The Deaf
* 0 NA values * total schools = 161 (5 NA
manually recoded)
Focus on odds ratio plot for levels by Tuesday–other plots by this weekend
Further cleaning needed. Multi-coded and non-coded data:
Paring down to district, charter, and independent
* charter = 58
* district = 77
* independent/other = 26
Table of Differences
Largest Differences
District vs. Charter Barchart
Charter vs. Independent
Independent vs. District Barchart
Smallest Differences
Individual Type Analysis
Charter vs. non-Charter Bar Chart
District vs. non-District Barchart
Independent vs. non-Independent Barchart
Variation Across All Types
Variability Between All School Types
Save plots:
Models
Race and SES
Predictors: number of students, black percent, hispanic percent, school type, FRPL percent, SWD percent, ELL percent
Plot found in outputs > type analysis
> model-race-ses.png. Code takes a long time to run, so
I don’t recommend pulling from this document.
Controlling for Aggregate BIPOC Population
The second model uses the same demographic variables, but rather than drawing from black and hispanic percentages, it uses an overall BIPOC student percentage (as calculated by non-white percent). I recognize this may not be super useful for reporting, but I was interested to see if there were any significant differences between majority-majority and majority-minority schools when racial demographics are not disaggregated.
Predictors: bipoc percent, school type, FRPL percent, SWD percent, ELL percent, number of students
Plot found in outputs > type analysis
> model-bipoc-ses.png. Code takes a long time to run, so
I don’t recommend pulling from this document.
Controlling for Locale
Predictors: bipoc percent, number of students, school type, school locale, FRPL percent
Plot found in outputs > type analysis
> model-type-locale.png. Code takes a long time to run,
so I don’t recommend pulling from this document.
Odds and Ends (7.6.22)
Outstanding questions to help build this out: How big are independent
schools (enrollment) compared to charters & districts?
* would you rather have this reported as averages within each type or
using bins?* What proportion of district schools are high schools? What
proportion of charters are HS? What proportion of independent schools
are HS? What proportion of charter schools are urban/suburban/rural,
vs. district and independent schools?
How big are independent schools (enrollment) compared to charters & districts?
I’d love to put a horizontal line in the violin plots indicating where the average is for each group so we can see the distribution relative to that, but I’m not sure how to do that with categorical data in this type of plot. Will do some digging if it seems worth it.